The question in the title is, in broad terms, one that arises when actions and emotions are assumed to have a certain pattern of similarities and differences; the question, broadly, is how to account for the differences compatibly with the similarities.

The similarities are that actions and emotions are, typically, reason-responsive (there can be good or bad reasons why someone, say, lifts her arm, or is angry), embodied (consisting, in part, in extra-neural bodily goings-on), and processes (occurrences that unfold in characteristic ways over time). These similarities mean, or so it will be argued, that emotions, no less than actions, typically can be described with progressive aspect as cases where someone is, say, trembling with fear, exploding with anger, exulting with joy, or in similar terms, bringing out their character of being embodied processes. As such, like actions, emotions are ‘a kind of doing,’ as Daniel Shargel and Jesse Prinz (2018:124) have recently described them. Moreover, like actions, the reasons to which emotions are responsive are deeply linked with desires, intentions, or ongoing projects: whether rain gives reason for joy or dismay has to do with what one wants or is up to.

The differences include that emotions typically are passive in a sense in which actions are not (‘it is obvious,’ remarks Ronald de Sousa, ‘to anyone not in the grip of a cocaine-inspired fantasy that there is something importantly right about the traditional view of emotions as passive’ (de Sousa, 2001: 112)). Relatedly, even if emotions are a kind of doing, they are, unlike actions, not typically cases of doing something intentionally. For example, someone who is trembling with fear, for a reason, is not typically doing so intentionally (which is not deny, of course, that emotions have intentionality, in the sense of directedness or aboutness).

The paper will not try to answer the titular question. Instead, as it were to prepare the ground for an attempt to answer it, I will argue it is hard to see how it could be answered within a widespread framework in the philosophy of action that has been dubbed the ‘Standard Conception of Action’ (cf. Schlosser, 2019: §2). That framework is a broad school, uniting causal theories of action in the mould of Donald Davidson (1980) and Michael Smith (2012) with such alternative accounts as those of G.E.M. Anscombe (1963), Jennifer Hornsby (2004, 2012), Michael Thompson (2008), and Julia Tanney (2013).Footnote 1 The framework has two key commitments. First, that the notion of action at the heart of the philosophy of action is to be approached in terms of the notion of doing something intentionally. Second, that the latter notion is closely linked with that of doing something for a reason, so as to imply what has come to be known as:

Anscombe’s Thesis If an agent φ for a reason, she φ intentionally.

Here and below, Greek letters hold the place for a description of what someone is or could be doing.Footnote 2 Now, if emotions are reason-responsive embodied processes, they will, as hinted, counterexemplify this. They will present myriad cases where someone is trembling with fear, exploding with anger, exulting with joy, etc. for a reason, but not intentionally. Otherwise put: if the Standard Conception were true, the titular question would have a false presupposition.

The main contentions of the paper, as the above intimates, are these. First, that the Standard Conception of Action is committed to rejecting the claim that emotions, even if typically not cases of doing something intentionally, are nevertheless, typically, reason-responsive, embodied processes. Second, that this claim about emotion has enough support and plausibility to cast serious doubt, given the first contention, on the Standard Conception.

The paper unfolds as follows. The next section introduces the Standard Conception. Section 2 fleshes out and supports the claim that emotions are, typically, reason-responsive embodied processes. Section 3 shows how emotions, as such, would commonly amount to counterexamples to the Standard Conception.

Section 4 contrasts this problem for the Standard Conception with the familiar problem of deviant causal chains (DCCs), notorious from Davidson’s (1980: 79) unnerved climber. It is worth highlighting a key difference already here: DCCs are a challenge not to the Standard Conception generally, but at most for causal theories of action that are subvarieties of that Conception.

Section 5 compares the problem emotions pose for the Standard Conception with a worry that mere beliefs and desires might be supposed to raise for it. Sections 6 and 7 answers rejoinders on behalf of the Standard Conception. Specifically, section 6 considers the claim that emotions, like beliefs or desires, but unlike actions, are not responses to ‘telic-instrumental’ reasons, i.e., roughly, to reasons having to do with the relevant type of act being a means to an end. Section 7 replies to the idea that emotions, like beliefs and desires, but unlike actions, are states, not processes. Section 8 concludes, noting some wider ramifications for current debates on action or emotion.

1 The Standard Conception of Action

The Standard Conception of Action has, as noted, two key commitments–it rests on two planks, if you wish (cf. Schlosser, 2019: §2).

The first plank concerns the notion of action at the heart of the philosophy of action. As a rough, initial gloss, this notion of action can be understood as one that paradigmatically includes human behaviours of interest to moral philosophy but excludes, e.g., such processes as breathing or blinking when merely automatic or reflexive. The first plank says that this notion of action is to be understood in terms of that of doing something intentionally (cf. e.g. Davidson, 1980: 43). Hornsby puts it so: ‘[t]he phenomenon of human agency can be caught in the first instance with the idea of someone’s doing something intentionally’ (Hornsby, 2004: §3, Hornsby's italics). The sense of ‘doing’ in play here is comparatively broad and unemphatic. Contrast a more stressed use of ‘doing’ on which someone who is merely breathing automatically is not really doing anything. Part of the point of the first plank is that we should not rely on such a stressed sense of ‘doing something’ to account for why a train of automatic breathing is not an action in the sense relevant to the philosophy of action. Rather, we should rely on the claim that automatic breathing is not case of doing something intentionally.

The second plank links the notion of doing something intentionally closely with that of doing it for a reason. The notion of ‘reason’ here is not that on which any old cause ipso facto is a reason, but one on which the status of something as a response to reasons goes with its being assessible for rationality, or reasonability, or in related, broadly normative-evaluative terms. Two subtly different formulations of the noted, close link may be distinguished. The first says that all and only things done intentionally are done for a reason (Davidson, 1980: 6, 83; Smith, 2003: 460). The second formulation, due to Anscombe (1963: 9), has it that what someone is doing is intentional just if a reason-requesting ‘Why?’ ‘has application’.Footnote 3 For our purposes we do not need to decide between them. Both formulations imply Anscombe’s Thesis, i.e., that an agent φ intentionally if she φ for a reason.

To get a sense of how the two planks may be fleshed out and put to work, it is useful to return to the chief architect of the Standard Conception, Anscombe; specifically, to her account of why someone who involuntarily gives sudden start when a face appears in a window is not acting intentionally (pp. 9–24, all unattributed references to Anscombe, 1963). Her account serves to anticipate how emotions pose a problem for the Standard Conception, as we shall see below.

Anscombe begins by setting aside two candidate accounts of why the jump in surprise is not an intentional action. The first invokes the idea, famously endorsed by Anscombe, that the applicability of the reason-requesting ‘Why?’ is subject to an epistemic requirement, viz. that agents know ‘without observation’ what they are doing and why, by which she means, at least, that agents know these things in an interestingly first-personal way, not merely by drawing inferences from evidence equally available to others. Anscombe grants however that the subject who suddenly takes fright might have such knowledge of his reaction. He might have an interestingly first-personal sort of awareness of how he is reacting and why, one he could voice by explaining ‘I saw a face in the window and it made me jump’ (p. 16).

The second candidate account proposes that the reaction here of giving a sudden start is not one of doing something or acting. Anscombe’s dismissal is instructive:

Why is giving a sudden start not an ‘action’ while sending for a taxi, or crossing the road, is one? The answer cannot be ‘Because the answer to the question “why?” may give a reason in the latter cases’, for the answer may ‘give a reason’ in the former cases too; and we cannot say ‘Ah, but not a reason for acting.’; we should be going round in circles. We need to find the difference between the two kinds of ‘reason’ without talking about ‘acting’, and if we do, perhaps we shall discover what is meant by ‘acting’ when it is said with this special emphasis. (p. 10, Anscombe’s emphases)

Here, Anscombe is clearly using ‘reason’ broadly for whatever is provided by an explanation; elsewhere she reserves it for what is provided by a certain special, rationalizing sort of explanation, corresponding to the ‘certain sense of “Why”’ said to be distinctively applicable to intentional actions. As noted, such a more restricted construal of ‘reason’ is adopted here. The cited passage reveals Anscombe’s commitment to the first plank of the Standard Conception. It would put the cart before the horse, she in effect argues, if we were to account for why jumps in surprise and its like are not intentional actions by appealing to a prior notion of acting, or doing something, that they allegedly fail to exemplify. Instead, the challenging cases should be rebutted by distinguishing the sort of explanation they are susceptible to from the rationalizing sort applicable to things done intentionally; their failure to be cases of acting or doing (said with ‘special emphasis’) is thereby to be elucidated.

Anscombe turns, therefore, to positing some characteristics of reason-giving explanations. In effect, she seeks to spell out how they go together with a certain normative-evaluative assessibility. Her characteristics are (labels mine):

(Evaluativeness) A good sign of reasons is that certain evaluative notions, such as of goodness or harm, are involved in them (pp. 21-3).

(Significance) The more a certain behaviour is ‘described as a response to something as having a significance that is dwelt on by the agent in his account, or as a response surrounded with thoughts and questions, the more inclined one would be to use the word “reason”’ (pp. 23-4, Anscombe’s italics).

(Argument-Underpinning Motive Link) ‘[I]t establishes something as a reason if one argues against it; not as when one says “Noises should not make you jump like that: hadn’t you better see a doctor?” but in such a way as to link it up with motives and intentions.’ (p. 24)

Anscombe also characterises a certain sub-set of reasons identified by explanations of intentional action as follows:

(Telic-Instrumentality) A mark of at least some reasons, viz. what she calls ‘further intentions’, is that they concern desired states of affairs conceived of by the subject as possibly brought about by the act of hers that we are seeking to explain (pp. 23, 35).

Anscombe certainly regards such telic-instrumental reasons, and reasoning with a means-to-an-end structure, as central to agency. She denies, though, that whenever someone acts intentionally, and the reason-requesting ‘Why’ has application, some of the agent’s reasons are telic-instrumental; their centrality is cashed out, rather, in terms of the idea that unless some action was for telic-instrumental reasons, we would not have the concept of intentional action that we do have (cf. pp. 30–4, and, for discussion of these points in Anscombe, Müller, 2011).

Anscombe suggests that accounts of involuntary reactions, along the lines of ‘I saw a face in the window and it made me jump’, will lack the above characteristics. Whether or not this is right for such a surprise reaction, I shall argue below it does not extend to emotions generally.

2 Emotions as Reason-Responsive, Embodied Processes

The view of emotions as, typically, processes that are at once embodied yet have enough of an intentional (in the sense of ‘being directed at something’), broadly cognitive aspect to be responses to reason and rationally assessible is perhaps first and foremost associated with Klaus Scherer’s (2000, 2009, 2022; Scherer & Moors, 2019) influential ‘Component Process Model’. It is worth stressing, however, that the view is not committed to the details of Scherer’s model and has several other notable proponents in psychology and philosophy.Footnote 4 Peter Goldie, a proponent in philosophy, nicely conveys the flavour of such a view:

An emotion—for example, John’s being angry or Jane’s being in love—is typically complex, episodic, dynamic, and structured. An emotion is complex in that it will typically involve many different elements: it involves episodes of emotional experience, including perceptions, thoughts, and feelings of various kinds, and bodily changes of various kinds; and it involves dispositions, including dispositions to experience further emotional episodes, to have further thoughts and feelings, and to behave in certain ways. Emotions are episodic and dynamic, in that, over time, the elements can come and go, and wax and wane, depending on all sorts of factors. (Goldie, 2000:12–13, cf. also Goldie, 2012:56-64)

I will now briefly explain and motivate reason-responsiveness, embodiment, processuality, and the claim that emotions commonly jointly exemplify these three marks.

Reason-responsiveness: To say an emotion, such as someone’s anger or sadness, is a reason-response is to say there are reasons for which she is angry or sad. The reasons may be good or bad. The reason-responsiveness of emotions is, then, closely linked with their being assessible as (ir)rational. The next section will show, more specifically, how Anscombe’s characteristics may apply.

If emotions embody some evaluation, appraisal, or judgement concerning the items towards which they are intentionally directed, it helps to account for how they can be reason-responsive. If anger, say, involves an assessment that one has been wronged or slighted, as Aristotle (cf. Rhetoric, 1378a21-1388b32) suggested and as more recent cognitivists have argued, we can understand how there can be good or bad reasons for being angry, e.g. because there can be good or bad reasons for construing the situation as such an offence.

However, the claim that emotions commonly are reason-responsive, embodied processes has only moderate commitments regarding cognitivism. It is consistent with a range of views except, on the one hand, a radical cognitivism that would reduce emotions to cognitive states, and, on the other, a radical anti-cognitivism taking all emotions to be so instinct-like to preclude reason-responsiveness. Those camps are uninfluential, though. Towards the cognitivist side, a theorist such as Solomon agrees in his later work that cognitive aspects of emotion are interwoven with and not be elevated at the expense of such bodily aspects as facial expressions and physiological activation (Solomon, 2003: 196–7). Proponents of views often contrasted with cognitivism, e.g. ‘basic emotion’ or ‘affect program’ theories, such as Ekman (2003, 2004) and Izard (2009), allow that emotions often involve complex, extended cognitive appraisals (cf. Scherer, 2022).

Embodiedness: Talk of an emotion as ‘embodied’ can have at least two senses. First, it can mean the emotion (partly) consists in bodily goings-on beyond the brain, say in trembling, or shivering; tensing up, or relaxing; breathing more rapidly, or more slowly; crying, smiling, blushing, or squirming. Second, it can mean the emotion (partly) consists in sensations, perceptions, or mental representations as of such bodily goings-on. The senses can be distinguished as ‘real’ and ‘experiential’ embodiment.Footnote 5

The claim that emotions commonly are experientially embodied is phenomenologically plausible. When someone is described as tense with fear, or bubbling with joy, say, it conveys a recognisable sense of what things are like for them. It is no easy task to imagine what things would be like for a subject were these bodily experiences to be absent yet the emotion itself left in place as it is. Imagining the bodily sensations away typically appear to detract from the very emotionality of the emotion. As Sartre (1939/1962: 49–50) suggested, these felt bodily upheavals seem ‘to represent the genuineness of the emotion’.

The claim that emotions commonly are really embodied finds support in correspondences among emotions and patterns of bodily activity. Joy, amusement, fear, and anger robustly correlate with distinctive facial expressions (cf., e.g., Ekman, 2003). Anger, fear, and joy go with sympathetic responses in the autonomic nervous system (ANS) (they are ‘high-arousal’ emotions), while sadness, contentment, and affection go with parasympathetic (they are ‘low-arousal’) (cf. Kreibig, 2010, Scherer & Moors 2019). It is disputed just how precisely specific emotion types can be mapped onto distinct bodily signatures. A review of 134 studies found however ‘considerable ANS response specificity in emotion when considering subtypes of distinct emotions’ (Kreibig, 2010). In any case, for emotions to have a vital bodily aspect that aspect need not all by itself serve to distinguish the type of emotion in question.

If emotions are really embodied, it strengthens the sense in which emotions can have distinctive functional roles. As Jenefer Robinson (2018:64) puts it, ‘emotions do not seem designed solely to track offenses and losses and so on. They also ready the organism to cope with these things.’ For example, in fear blood tends to be redistributed from the viscera to the large muscles (cf. Clore, 1994:110–1); if this is part of fear, we understand how it prepares for flight. Anger has its typical facial and postural expressions, e.g., solid stance, or clenched fists, which serve to prepare for attack (cf. Frijda, 1986:11–12).

If emotions are really embodied, it also helps underwrite how they can have distinctive socio-communicative roles. It provides a robust sense in which we can see the joy in someone’s face or hear the anger in their voice (cf., e.g., Husserl, 1901/2001:190). In seeing someone’s frown, hearing their raised voice, etc., we may be beholding, not mere extrinsic signs or symptoms of their anger, but part of that in which their anger consists (cf., e.g., Green, 2010).

It has been argued, inter alia on the above-mentioned grounds, that taking emotions to be partly constituted by bodily processes enhances the prospects of defining emotions generically, as against moods, thoughts or perceptual experiences, and certain specific types of emotion, such as anger versus fear, in a way that carves at the psychological joints (cf., e.g., Mulligan & Scherer, 2012; Scarantino & Griffiths, 2011).

When, in what follows, I say emotions are ‘embodied’, I mean they are both experientially and really embodied (unless otherwise clear from context).

Processuality: A key motivation for thinking of emotions as processes is the thought that emotions are neither just cognitions, nor mere bodily upheavals, nor simply sensations thereof – but neither are they mere collections of such mental or bodily ingredients. These facets of the emotion interact, in distinctive, coordinated ways over time. Two representative quotes, supplementing that from Goldie above, from, respectively, Robinson and Phoebe Ellsworth:

Emotions are not instantaneous but take time, and different components in the emotion process are typically on different timelines. In my view, when philosophers claim that emotions are appraisals or attitudes or whatever, they are simply focusing on what they take to be the essential feature of an emotion. But this is a mistake. Emotions are essentially processes involving a number of components (Robinson, 2018:52, her emphases)

Interpretation, subjective feeling, visceral and motor responses are all processes, with time courses of their own. (…) The interpretation develops over time, and so does the feeling, in a continuously interactive sequence, often a very rapid one. (Ellsworth, 1994:227)

For example, a perception of a growling Rottweiler and appraisal of it as threatening may set in train a chain of interacting physiological reactions, feelings, facial and postural expressions, and action tendencies, where these in turn are apt to readjust one’s evolving appraisal of the dog.

A standard definition of a process is as an ‘actual or possible occurrence that consists of an integrated series of connected developments unfolding in programmatic coordination: an orchestrated series of occurrences that are systematically linked to one another either causally or functionally.’ (Rescher, 2000:22) Emotions would seem to meet this description. The various facets of the emotion are just such ‘connected developments’, linked causally as well as functionally, subject to a form of ‘programmatic coordination’ thanks, notably, to the role of cognitive appraisals, which not only tend to elicit these developments but have a role in monitoring them as they unfold (Robinson, 2018; Scherer, 2000, 2009).

While this dynamic view of emotion is not motivated principally on phenomenological grounds, it is, I think, phenomenologically apt. Emotions commonly seem to boil up, peak, and then subside. Some emotion types seem to go with felt developmental trajectories of more specific types, as when a sudden apprehension is described as a sinking feeling, or a bout of anger as feeling as if one is to explode. There is a structural parallel here to other experiential processes, e.g. that of hearing tension build up and then being released in a chord progression.

The rationality of emotions, and the reasons to which they are responsive when broadly apt, also fit construing emotions as processes. The news that one has been wronged may make anger apt. Even a serious wrong would not, however, typically make it apt for one’s anger to rage on indefinitely. As anger works its way through one’s system, the reasons for one to be angry are as it were consumed in the process (cf. Na'aman, 2021). There is a contrast here with reasons for belief, which are not consumed in the act of believing: reasons for believing you were wronged remain in force, at least until countervailing evidence emerges. The supposition that anger by its nature is a process one goes through, with a characteristic pattern of boiling up, various subsequent stages of simmering, before cooling off, helps make sense of why the reasons should have this character.

Commonly reason-responsive, embodied, and processual: For the purpose of bringing out a tension with the Standard Conception of Action, there is no need to assume that all emotions have these three marks; it is enough (or more than enough) that they commonly do.Footnote 6 Some emotions may be, say, mental states rather than processes; some may be too low-level be reason-responsive; and some may lack bodily components.

The question how emotions commonly could jointly exemplify the three marks may not be pressing when it comes to the combination of processuality with embodiedness, since the latter, e.g. in the guise of shivering, breathing rapidly, or squirming, often naturally are understood as processes. The joint exemplification of processuality-cum-embodiedness with reason-responsiveness may seem puzzling, though: How can an embodied process be a rationally assessible response to reasons? This is undoubtedly a good question. Emotions-as-processes theorists have however offered suggestions towards answering it, emphasising the initiating and monitoring role of cognitive appraisals in the overall emotion process (Robinson, 2018; Scherer, 2000, 2009). In this dialectical context we can, moreover, afford to rest content with noting that there is a powerful reason to think embodied processes can be responses to reasons. Paradigm intentional actions, e.g., brewing beer, are plausibly embodied processes, and it is a truism that they can be responses to reason: one can be brewing beer for a reason.

3 Emotions as Cases of Doing Something for a Reason, but not Intentionally

Anscombe’s account of why an involuntary jump in surprise is not something done intentionally allows, as we saw, that it is a case of doing something (and even one where the subject may well have a first-personal awareness of how she is reacting and why). The key point, for her, is that the reaction is not a response to reasons.

If the above is on the right lines, this account will not generalise to explain why emotions are not things done intentionally.Footnote 7 If emotions are embodied processes, they may aptly be described in ways bringing out their status as such, e.g. as in:

(1)

a. She is swelling with pride

b. He was trembling with fear

c. She was revelling in the thought of her new-found success

d. She will be fuming with anger

e. She is exulting with joy

These descriptions highlight the bodily aspect of the emotions in question. By using the progressive form (‘is swelling’, ‘was revelling’, etc.), they convey a sense of the emotions as unfolding processes (cf. Section 7 below). If someone who is breathing automatically, or jumping in surprise, qualify as doing something, under the first plank of the Standard Conception, someone exulting with joy or trembling with fear should too. What (1a–e) describe may moreover be responses to reasons.

For example, consider Liz, a five-year old, who for days anxiously has been awaiting her cousin’s arrival at the cottage, and who suddenly exults with joy, smiling raptly, her eyes widening, her pulse quickening. Her exultant joy here is not something she puts on intentionally; it is not an intentional action (although, again, it has intentionality in the sense of being about or directed at something). She has no intentions or ongoing projects or is making no attempts which she is regarding her way of reacting as a means to fulfil or a way of fulfilling. Yet we can ask: Why is she reacting as she does here? An answer is given when she exclaims: ‘Look! Uncle’s car is coming up the driveway!’ What Liz offers here is not (merely) some brutely causal influence on her emotion. It is to be contrasted with an account of her reaction that adverts, say, to her having underslept the last few days. What she offers is something that shows up within her perspective on the world, and which, within that perspective, figures as a reason for her reaction. Anscombe’s own features suggest so much:

(Evaluative) When Liz exults with joy, spotting uncle’s car coming up the driveway, her response characteristically reflects an evaluation of that event as meaning that something good is about to happen.

(Significance) Liz joy, beholding uncle’s car, not untypically involves her dwelling upon various further things signified by it: how it indicates the arrival of her cousin, the end of her loneliness, the beginning of untold adventures playing in the meadows, etc.

(Argument-Underpinning Motive Link) The explanation underpins discussion of how good a reason the indicated explaining circumstances are for the reaction explained, discussions that often would link up with the motives and intentions of the subject whose reaction it is (recall, again, how rain may give reason for either joy or dismay, depending on what one is up to). In this case, Liz’s mother may have to disappoint her: ‘Unfortunately, your cousin has, I heard, broken her leg. You won’t be able to play in the meadows, I’m afraid.’

It is true that the account of Liz’s reaction given by her exclamation lacks the feature of Telic-Instrumentality: it does not identify a desired states of affairs conceived by her as possibly brought about by her joyful reaction. However, as we saw, this would not, by Anscombe’s lights at least, show Liz’s reaction not to be a response to reasons (show it not to be a case φ-ing for a reason), in the sense relevant to the Standard Conception.

There is nothing out-of-the-ordinary about Liz’s case. Similar, unexceptional accounts could be given for myriad everyday emotional reactions, yielding scores of cases where someone is doing something – swelling with pride, freezing up with fear, or what have you – for a reason, but not intentionally.Footnote 8

4 A Comparison with Deviant Causal Chains

In a broad sense, a deviant causal chain (DCC) in the domain of agency is any case where agents do something, causally explained by their beliefs, desires, intentions, or ongoing projects, without doing it intentionally. Thus Davidson’s (1980: 79) climber wants to get rid of the weight of his partner, realises he can do so by letting go of the rope, and is so unnerved by this that he is caused to let go of the rope, not doing so intentionally. Liz’s case, and myriad emotional reactions like it, are DCCs in this broad and rather technical sense, if explanations of these reactions in terms of the reasons to which they are responsive are causal explanations. Even so, however, they would be very special cases of DCCs.

First, the stock assumption about DCCs is that the agent is not doing what she is doing for a reason. Indeed, when causal theorists seek to overcome the problem of DCCs and formulate a sufficient condition for intentional action in causal terms, it is sometimes suggested that what they need to ensure is precisely that such-and-such mental states not only cause what the agent is doing but cause it ‘qua reasons’. Marcus Schlosser writes:

The reason why the causal pathway in the climber example is deviant seems to be that the movement is merely caused by the reason states – the movement is not a response to the reasons qua reasons. (Schlosser, 2007: 191)

Liz’s exultant joy is however not, or not merely, caused by such and such belief and desires of hers, but a response to them sensitive to their status as making her jubilation eminently reasonable.

Second, while standard-issue DCCs are set out in explicitly causal terms, this is not true of our case of Liz, and need not be true of other accounts of emotions like it. Indeed, the characteristics of reasons invoked in our argument that Liz’s reaction is subject to a reason-giving explanation derive from Anscombe, who not only opposed a causal view of such explanations but adverted to the indicated features precisely to contrast them with causal explanations (cf., e.g., Tanney, 2013: 103–88). In so far as Liz’s exultation and emotional reactions more generally present a challenge to the Standard Conception, the challenge is not contingent on presuming a causal view of reasons-invoking explanation. While causal theorists of action could, adopting the broad, somewhat technical sense above, class them as special cases of DCCs, non-causal theorists should not view them as DCCs.

Third, a common suggestion about DCCs is that there is something fluky or fortuitous, and therefore deviant, about how such-and-such causing attitudes led the agent to do just so-and-so.Footnote 9 Again, this does not apply to Liz’s case – there is nothing deviant about it (so qualifies as a DCC at most in a technical sense). Her jubilant reaction to her situation, as she perceives it, may be reliably expressive of her well-honed sensibility to things at stake for her, affectively, in the situation at hand.

5 Are Beliefs or Desires no Less a Problem for the Standard Conception?

Someone who have followed the argument so far, may be tempted to respond as follows, which I dub:

The Belief/Desire Worry

‘Why enter the murky terrain of emotion to find trouble for the Standard Conception? If trouble can be found there, could it not more easily and no less effectively be found with the familiar attitudes of belief or desire? People believe things, and desire things, for reasons. Even those who hold that we are in some sense responsible for our beliefs do not usually go so far as to claim that we believe that p intentionally whenever we believe that p for a reason. It would be widely agreed that one can want something for a reason, even if one’s wanting it is an entirely passive, involuntary matter. Moreover, the verb “does” was supposed to be understood in a rather thin and broad way for the purposes of the Standard Conception. Would not “doing” then subsume believing and wanting? At least, we say things like “Peter believes in an afterlife, and Paul does; Paul wants us to be pious, and Peter does.”’

Does this worry present any less of a problem for the Standard Conception than emotions? Indeed, does it not present essentially the same problem? If so, doubts may arise if we have not been misconstruing the Standard Conception. Its many friends may or may not have been neglecting the worries over emotion that Anscombe dimly glimpsed. Surely, however, they have been aware of the commonplace that one can want or believe something for a reason, without doing so intentionally. Would it not be uncharitable to construe these writers as erecting their views upon assumptions incompatible with that commonplace? Yet, if their assumptions, properly understood, are compatible with that commonplace, and emotions present essentially the same problem, we can be confident in setting aside the challenge from emotion.

I will here remain neutral on just how worrisome the belief/desire worry ultimately should be for the Standard Conception. I deny, though, that it presents essentially the same problem as emotions. For one thing, beliefs and desires do not constitutively involve the body as emotions and actions typically do. At least, bodily action has pervasively been the paradigm of intentional action in philosophy of action.Footnote 10 Though there is also mental action, it has been argued it differs deeply from bodily action (cf., e.g., O’Shaughnessy 2009; Proust, 2013). Thus, at least prima facie, one option open to the friend of the Standard Conception, to allay the belief/desire worry, is to clarify that the conception applies, in the first instance, to bodily doings. The next two sections will look at two other responses a friend of the Standard Conception may venture. They may have some promise when it comes to tackling the belief/desire worry. I shall argue that they are unsuccessful, however, when it comes to emotion.

6 Responsiveness to Telic-Instrumental Reasons?

When people react emotionally, the reasons for which they react as they do rarely seem to have the feature of Telic-Instrumentality: the reasons rarely take the form of a desired state of affairs, conceived by the subject as possibly brought about by the emotion we want to explain. Though Anscombe denies that such Telic-Instrumental reasons apply whenever someone acts for reasons (and intentionally), other proponents of the Standard Conception can be understood to hold that they do. Davidson, for example, writes:

Whenever someone does something for a reason … he can be characterized as (a) having some sort of pro attitude toward actions of a certain kind, and (b) believing (or knowing, perceiving, noticing, remembering) that his action is of that kind. (Davidson, 1980: 3-4)

This claim, or a slight weakening of it, can be put as:

The Telic-Instrumental Restriction An agent φ for a reason, in the sense relevant to the Standard Conception, only if her φ-ing is rationalised by her wanting, intending, or having an ongoing project to ξ, and her regarding φ-ing as likely being a way of or conducive to ξ-ing.

Several philosophers of action beside Davidson seem to have assumed this restriction. For example, Michael Smith (2003, 2012) relies on it, in conjunction with the claim that agents do something intentionally iff they do it for a reason, to argue for the Humean claim that the reason for which someone acts always partially consist in a desire. Hursthouse (1991) assumes the restriction in arguing that some behaviours expressive of emotion, such as smashing crockery out of anger, are intentional and yet not performed for a reason, since the agent does not conceive of the behaviour as a means to something she wants.

The Telic-Instrumental Restriction, we may observe, promises to help with addressing the belief/desire worry. Arguably, we cannot believe things, e.g. that we are fantastic, just on the grounds that we believe believing it will lead to things we want, e.g. behaving confidently. When we want something, e.g. money, for a reason, this arguably has to be because we take the thing we want, i.e. money, to be something that enables something else we want or value, e.g. buying something, rather than because we take wanting money to lead to something wanted. Of course, you can want to want money, because you think wanting money encourages, say, getting money. Again, however, it is what you want (i.e. wanting money) and not wanting it (i.e. wanting to want money) that you believe to lead to something wanted or valued (cf., e.g., Gibbons, 2009). This is not yet to show, of course, that the Telic-Instrumental Restriction ultimately overcomes the belief/desire worry, only to suggest it has some promise.

However, that restriction does not handle all problems posed by emotion. The restriction fails because a great many emotional reactions are ‘higher-order’, in that they are informed by an awareness of actual or possible emotions in ourselves or others, or of how we or others may or may not be affected by emotions. I might revel in the news an embarrassment befalling my adversary, and then, in turn, become ashamed of my enjoyment. Sometimes emotions feed on themselves in this way. News of actual or possible joys might give grounds for joy; news of likely desirable effects of joy might too. Fury might cause certain desired reactions in others, e.g. that they at long last take my concerns seriously, a fact that itself may be infuriating. So I may, quite unintentionally, get infuriated, for the reason that, as it now dawns on me, getting infuriated is the one and only way of getting what I want.

Or consider this happier example. Face to face with my sweetheart, I dearly want her to smile her gorgeous smile, and set about to make her do so. Knowing her sweetness, and how perceptive she is regarding my emotions, wants, and intentions, it occurs to me that, if I were to break out in joy, she would smile precisely that gorgeous way. The sudden realization of this heart-warming, charming fact may be just what leads me to break out in joy, quite involuntarily. My realization of the charming fact, given my project in this case, rationalizes the reaction. It reveals something in the light of which it makes excellent sense to revel in joy. When I react emotionally as I do, it is no mere fluke that I do, but expressive of my emotional sensibilities – of my being perceptive of and attuned to something that matters to me affectively in the situation as hand.

True, it might well be misleading, if my sweetheart asked me why I reacted as I did, to reply merely ‘I wanted to make you smile’. For, on broadly Gricean grounds of quantity and manner, this might well suggest, falsely, I reacted as I did in order to make her smile (so intentionally). The reply would not be false, though. If I had been indifferent or averse to making her smile, I would not have reacted as I did. My desire, moreover, is no merely or brutely causal influence, but intimately linked to the aptness and, in broad terms, reasonability of my reaction. The reply could be expanded to remove the false suggestion: ‘Well, you know, I really wanted to make you smile, and then it struck me that if I were to break out in joy, you would smile gorgeously – and this heart-warming realisation made do so, without meaning to.’Footnote 11

The just-outlined case has visible parallels with such classic DCCs as Davidson’s climber. The three contrasts with standard-issue DCCs noted in §4 still apply, though. My delighted reaction is not (merely) caused by my realization of the charming fact, along with my heartfelt desires and projects, but sensitive to them in their capacity of revealing such a reaction as an apt one in this case. No explicit controversial causal assumptions are needed to articulate the case. An anti-causalist may agree our case is one where certain attitudes or projects of mine make it intelligible and reasonable that I react as I do, while denying that they cause the reaction. My reaction, moreover, is not fortuitous or fluky, but reliably expressive of my emotional sensibilities.

7 States, not Processes?

Sympathisers with the Standard Conception are likely to reply to the challenge from emotion, and indeed to the belief/desire worry, that emotions, beliefs, and desires are mental states, whereas actions are dynamic entities. States obtain at or through an interval of time, while dynamic entities such as events or processes occur, happen, or unfold at or over times. The notion of ‘doing something’ operative in the Standard Conception is perhaps capacious but should not be construed so capaciously as to subsume states.

An influential reason to consider emotions, beliefs, and desires states turns on the lexical aspect of verb phrases (VPs) by which they are attributed.Footnote 12 Consider:

(2)

a. He is happy/sad/afraid/angry.

b. She loves/hates him.

c. She believes that it is snowing.

d. He wants to go to the cinema.

These VPs have various marks by which they have been assigned to a category of state-VPs, contrasting with VPs in dynamic categories. They are ungrammatical or infelicitous in the progressive:

(3)

a.? He is being happy/sad/afraid/angry.

b.? She is loving/hating him

c.? She is believing that it is snowing

d.? He is wanting to go to the cinema.

Contrast here with paradigm attributions of action:

(4)

a. He is lifting his arm.

b. She is building a house.

The VPs in (2), unlike those in (4), are also unhappy as ‘bare infinitive’ complements in perceptual reports (cf. Maienborn, 2019: 31–42):

(5)

a.? She saw him be happy/sad/afraid/angry.

b.? They saw her love/hate him.

c.? We heard him want to go to the cinema.

d. He saw him lift his arm.

e. We heard her build a house.

Again, the VPs in (2) contrast with dynamic VPs like those in (4) in their susceptibility to combining with a succeeding event anaphor (cf. Maienborn, 2019: 57–59):

(6)

a. ? He was happy/sad/afraid. This happened while…

b. ? He wanted to go to the cinema. This happened while…

c. He lifted his arm. This happened while…

d. She built a house. This happened while…

The VPs in (2) are, inter alia on these grounds, often offered as paradigms of state-VPs by semanticists (cf., e.g., Bhatt & Pancheva, 2005: 9).

However, even if the VPs in (2) are state attributions, to gauge what this means for the Standard Conception we need to ask what significance attributions of the forms in (2) should have for our conception of the emotions or attitudes in question.

For beliefs and desires, sentences of the form ‘S believes that P’ and ‘S wants to φ’ may well be canonical ways of attributing these attitudes. Assuming that these sentences attribute states of belief or desire, it is not clear what reason there is also to posit doxastic or conative processes, where the latter combine the properties of being responsive to reason and non-intentionally executed, thereby making problems for the Standard Conception. Therefore, the state-VP character of (2c, d) has some tendency to allay the belief/desire worry.

For emotions, however, the canonical status of state-VP forms such as ‘S is happy/sad/afraid/…’ is more open to doubt. In (1), above, we offered examples of emotional attributions happily taking the progressive. These and related attributions also commonly exhibit other marks of dynamic status:

(7)

a. He was shivering with fear.

 

b. They saw her fume with anger.

 

c. She exploded with rage. This happened while…

These are not contrived or unusual ways of speaking of emotion, but arguably vital means of conveying information thereof, especially when we want to impart a lively sense of their very emotionality.Footnote 13 It may be complained that these ways of speaking are metaphoric, metonymic, or otherwise indirect, colourful ways of getting at what would be more directly and revealingly attributed by means of ‘S is (very) afraid/angry/happy’. However, it would obviously beg the question at stake here to support this complaint by appeal to the idea that emotions only or primarily take the form of states, if that idea is not motivated on other, non-linguistic grounds. Emotions-as-processes theorists could, of course, press the contrary complaint, viz. that preoccupation with such state-VP forms as ‘is afraid/angry/…’ hampers understanding. Scherer has argued along just these lines (see also Frijda, 2008: 73–4):

[A] true science of emotion (or "affective science") needs to keep its distance from the cherished habit of anchoring all thinking and research about emotion around a convenient set of apparently natural categories of affective states as provided by the respective languages… It is a valid, even fascinating exercise in semantics to unpack the meaning structures of emotion terms in a language… However, this should be but one facet in the multifaceted enterprise of understanding the nature of emotion, not the royal road.

… [M] any of the established, paradigm-driven research traditions use concepts or research procedures that are inconsistent with the evidence that is accumulating with respect to the phenomena. This is particularly true with respect to treating, despite occasional verbal declarations to the contrary, emotions as steady states rather than processes. Again, this problem has its roots in the domination of emotion research by verbal labels that, by their very nature, refer to stable states. (Scherer, 2000: 72–3).

In sum: some familiar ways of talking about emotions invite conceiving of them as states, but other ways of describing them encourage thinking of them as processes; which of these descriptions portray emotional life more revealingly is unlikely to be decided on linguistic grounds.

Now, even if it is conceded that attributions of the form ‘S is afraid/angry/happy’ cannot necessarily be assumed to be canonical, an emotions-as-states theorist may rightly insist that they are very often true. A plausible semantics will take their truth to depend on instantiation of emotional states. No-one would want to think there are two families of emotions, states and processes, that float free of each other, or compete to play the role of emotions. Therefore, it is advisable to identify emotions with these states to which we are anyhow committed, or so it may be argued.

This brief argument is unbalanced, neglecting to note that there is empirical-cum-theoretical evidence, indicated in Sect. 2, as well as such putatively true common-sensical remarks as, e.g., (7a-c), that indicate that there are processes that are, if not identical with, at least intimately related to emotions. The question how to reconcile evidence for emotional processes with evidence for emotional states is anyone’s problem in this area. Moreover, the emotions-as-processes theorist has various options when it comes to accounting for the truth of sentences of the form ‘S is afraid/angry/happy’, without encouraging a problematic double-counting or explanatory competition. One such option (not the only one) is to adopt a proposal of Claudia Maienborn’s (2019). She argues state-VPs such as ‘S is afraid/angry/…’ refer to what she dubs ‘Kimian states’, drawing here on Jaegwon Kim’s notion of events as temporally bound property exemplifications. Specifically, Kimian states are for her ‘abstract objects for the exemplification of a property P at a holder x and a time t.’ (2019: 46) For the case of, say, ‘S is angry’ the exemplified property, P, in the Kimian state referred to, may simply be the property of undergoing a process of anger. On this analysis, admitting the truth of such state-VP emotional attributions does not commit one to any concrete emotional states; emotional states are abstracta encoding the undergoing of emotional processes.

Recently, Matthew Soteriou (2018) has proposed another way of reconciling the evidence for emotional states and for emotional processes. His proposal has close affinities with process views such as Scherer’s. Like these, he regards an emotion as a unitary phenomenon having bodily and experiential processes as constituents. He however identifies this unitary phenomenon that these processes somehow make up and constitute as a mental state rather than a process.

Soteriou argues his conception safeguards the explanatory integrity of emotions. He inveighs against Goldie’s view that Goldie does not do enough to account for how emotions, as opposed to their component (sub-)processes, can do explanatory work. This challenge, of accounting for the causal-explanatory integrity of emotions, is no doubt an important one. Doing justice to it is beyond the scope of this paper. Suffice it to note that Soteriou gives no general reason to think emotions-as-processes theorists cannot meet it. There are various accounts, alternative or supplementary to Goldie’s, of what integrates the component processes in an emotion (cf., e.g., Lewis, 2005; Scherer, 2000). In any case, it is plausible to think that complex processes, rich in sub-processes, can do important causal-explanatory work, as witness such biological or chemical processes as metabolism, photosynthesis, etc. If emotions are properly integrated processes, why should their status as processes rather than states be held against their causal-explanatory credentials?

Besides, as Soteriou’s himself suggests, emotions would on his view often be highly un-static. He argues his view ‘accommodates something that Goldie was keen to emphasise: the idea that there can be characteristic patterns to the unfolding of occurrences–events/processes which are to be thought of as constituents of the emotion’ (2018: 89, his emphasis). These characteristically shaped modes of unfolding may specify, for Soteriou, the nature of the emotion. So, the emotion would not be static but develop in characteristic ways over time, and not thanks to changing extrinsic relationships but of its intrinsic nature. If states are to be contrasted with events and processes as static items versus dynamic (as Smith (2018) for one suggests), emotions fall, then, with the latter. Perhaps that way of contrasting states with events/processes is too simplistic; perhaps states can be significantly dynamic. Adopting such a conception comes with a drawback, however, for the theorist keen to support the Standard Conception along the lines indicated at the start of this section, i.e., by restricting doings to dynamic items, as the restriction would no longer exclude states.

8 Implications

The Standard Conception of Action holds that action is to be understood in terms of the idea of someone’s doing something intentionally, and links doing something intentionally closely with doing it for a reason, so as to imply Anscombe’s Thesis that things done for a reason are done intentionally. The conception underlies a widespread approach to agency, seeking its nature in a rationalizing relationship between certain doings, on the one hand, and a combination of beliefs, desires, intentions, or overarching projects, on the other. The possibility of deviant causal chains has long been recognised as a thorny problem for causal accounts of this rationalizing relationship. We have seen that, on the plausible view of emotions as reason-responsive embodied processes, emotions will often aptly be understood as cases where someone, say, shivers with fear, or exults with joy, for a reason, but not intentionally. Unlike the problem of deviant causal chains, this problem arises independently of the choice between causal or non-causal interpretations of the rationalising relationship at heart of the Standard Conception.

Beside challenging the Standard Conception generally, the thoughts on emotions offered in this paper also bear on some other, more specific debates on agency. One debate concerns the diagnosis of causal deviance. Let’s suppose, with causal theorists, that reason-giving explanations are causal. Then our counterexamples, like Liz’s exaltation in Sect. 3 above, form a subvariety of DCCs in the broad sense noted in Sect. 5. A causal account of action that handles DCCs must, then, ipso facto handle these cases. As we saw, it has been suggested that deviance is a matter of being causally sensitive to some mental states but not ‘qua reasons’ (Schlosser, 2007: 191), or that there is something fluky about how those states led to the reaction in question (Bishop, 1989:148–9, Aguilar, 2012: 3). Neither of these diagnoses will typically be apt to explain why ordinary, rationally apt reactions of freezing with fear, exulting with joy, etc. are not cases of acting intentionally.

Another debate on which this paper’s thoughts upon emotion bear concerns the merits of an ‘intentional action first’ view of agency, as has been proposed by Williamson (2017) and Levy (2013), by analogy with Williamson’s ‘knowledge first’ view. Williamson and Levy argue there are grounds for pessimism about the prospects for an informative sufficient condition for intentional action, in terms that do not use that notion, just as, Williamson argues, the Gettier problem favours pessimism about such a condition for knowledge. Their analogy of the Gettier problem for the case of action is the problem of DCCs. The drift of this paper supports their argument here. It shows that a cousin of the problem of DCCs arises even for non-causal accounts within the Standard Conception. In addition, as suggested in the last paragraph, it deepens the problem of deviance for causal accounts within that framework.

The emphasis of this paper has been on drawing lessons for the philosophy of action from emotion research. Charting these implications would be of interest, however, even if, contrary to the drift of this paper, the Standard Conception ultimately should stand. In that case, our contention that the Standard Conception is committed to rejecting the view of emotions as reason-responsive, embodied processes could be used instead to ‘go modens tollens’, to show that the various empirical and theoretical considerations marshalled in favor of that view must be reconsidered. A broader message of the paper would, then, still be borne out, viz. that philosophical work on action and on emotion would benefit from increased attention to the similarities and differences between these domains of mind.